Journal of Biogeography, 29, 387–394 Island biogeography and metapopulation dynamics of Bahamian ants Lloyd W. Morrison Section of Evolution and Ecology, Division of Biological Sciences, and Center for Population Biology, University of California, Davis, CA 95616, USA Abstract Aim I examined the island biogeography and metapopulation dynamics of ants inhabiting two archipelagos of small Bahamian islands. Of particular interest were measurement and comparison of turnover rates, examination of variation in relative population abundances, and the effect of a hurricane force disturbance on the ant fauna of these small islands. Location Archipelagos of small islands in the central Exumas and near the northeast coast of Andros, Bahamas. Methods Ants occupying small islands were surveyed using tuna and honey baits. I surveyed ninety-three islands in the Exumas in 1998 and fifty-eight islands at Andros in 1999, to compare with earlier surveys in both regions. The proportions of baits occupied were used as a measure of relative population abundance. A subset of seventeen small islands in the Exumas was surveyed in 1999 in the aftermath of Hurricane Floyd. ) Results Mean annual relative turnover rates were low: <2.5% year 1 on a per island ) basis, and <7% year 1 on a per species basis. Rates of immigration and extinction were similar, although immigrations exceeded extinctions in some comparisons. Relative population abundances of the two most common ant species varied inversely with each other. One species revealed a strong positive correlation with recent rainfall, whereas another varied strongly inversely. No extinctions of ants occurred on the seventeen small islands surveyed after Hurricane Floyd. Main conclusions Ants were found to be ubiquitous in this system, occurring on almost all vegetated islands. Ant populations were persistent over the period of study, and species rarely became extinct or colonized islands. The few instances of turnover observed appeared to occur randomly with respect to physical island characteristics. The correlational data suggest an interaction of interspecific competition and precipitation affect relative population abundances. Ants were found to be resistant to hurricane-force disturbances. In the short term (one decade), the ant fauna of these islands appears to be in a state of static equilibrium, although non-equilibrial dynamics may better characterize the system over longer time periods (several decades). Keywords Bahamas, Formicidae, hurricane, species turnover. Brown & Lomolino, 1989). This body of theory has inspired INTRODUCTION many investigations of insular biotas (reviewed in Williamson, The equilibrium theory of island biogeography (Macarthur & 1981; Rosenzweig, 1995; Brown & Lomolino, 1998; Whit- Wilson, 1963, 1967) provided a pioneering quantitative taker, 1998), although numerous criticisms have been levied approach to the study of fragmented populations (but see against the simple equilibrium model (e.g. Gilbert, 1980; Williamson, 1989; Whittaker, 1998). The related concept of the metapopulation (Levins, 1969, Correspondence: Center for Medical, Agricultural and Veterinary Entomol- ogy, USDA-ARS, PO Box 14565, Gainesville, FL 32604, USA. 1970) provided an alternate theoretical underpinning for E-mail: [email protected]fl.edu investigations of insular populations, focusing on the Ó 2002 Blackwell Science Ltd 388 L. W. Morrison archipelago-wide immigration-extinction dynamics of one or In the Exumas, I surveyed ninety-three vegetated islands in a few species, rather than species number on a single island. May 1998. All islands had been surveyed last in May 1994, In the last decade, a paradigm shift in the study of and most had been surveyed at annual intervals in the four fragmented populations has occurred, in which metapopu- preceding years (see Morrison, 1998a). At Andros, I surveyed lation theory (e.g. Gilpin & Hanski, 1991; Hanski & Gilpin, a total of fifty-eight islands in April – May 1999. Forty-eight 1997) has replaced the equilibrium theory of island bioge- islands were vegetated, and ten lacked vegetation. Forty-six ography as the dominant theoretical approach (Hanski & of the vegetated islands had been surveyed previously for Simberloff, 1997). A complementary analysis (Ouborg, ants in 1990. 1993; Morrison, 1997a, 1998a; Vidal et al., 2000), how- On 14, September 1999, Hurricane Floyd, a strong ever, may elucidate more about fragmented populations than category 4 storm, passed c.100 km NE of the Exumas either approach alone. archipelago. Conditions in the study area approximated a Recently, a call for a new paradigm of island biogeogra- class 2 hurricane. The rise in sea level in the immediate phy has been issued (Global Ecology & Biogeography, vicinity of the islands as a result of the hurricane was 2000). A crucial question in modern island biogeography estimated to be at least 1.5 m, based upon debris washed up theory is whether islands ever actually reach a state of on nearby islands and interviews with natives. During 25–30 equilibrium, or whether the frequency of disturbance is too September, 1999, I re-surveyed seventeen of the smallest and great (Whittaker, 1995, 2000). A major challenge to lowest vegetated islands in the Exumas, to compare with resolving this issue is obtaining data sets of sufficient surveys conducted in May 1998. The vegetated areas of the temporal and spatial scales. With the possible exception of islands ranged from 0.2 to 474.6 m2, and elevations ranged butterfly metapopulations, which have been relatively well from 0.55 to 2.74 m. studied (Thomas & Hanski, 1997), data on invertebrate persistence in very small habitat fragments remains scarce Survey methodology (Abensperg-Traun & Smith, 1999). Furthermore, a large proportion of existing studies have focused on species that Baits of tuna and honey were used to attract the diurnally exhibit relatively high rates of turnover (e.g. Schoener, 1983, active ant species inhabiting these islands. Baits were placed 1991; Bengtsson, 1991). uniformly over each island, the number of baits being Previously, I reported results of a 5-year study of the proportional to the log(area) of the island, and ranging from ants occupying an archipelago of small islands in the a minimum of six on the smallest to a maximum of fifty on central Bahamas (Morrison, 1998a). Here I present addi- the largest. Approximately 1 g of finely ground tuna in tional results from subsequent work on small Bahamian vegetable oil (to provide proteins and lipids) and 1 g of islands across two archipelagos, from a data set that now honey (to provide carbohydrates) were placed in Petri dishes spans almost a decade. This paper addresses the following 58 mm in diameter. The Petri dishes (painted white) aspects of this insular ant fauna: (1) How frequently do prevented the oil from soaking into the ground and ants turn over on these small islands, and how does this facilitated ant identification at baits. The sides of the Petri vary spatially and temporally? (2) How do relative dishes (approximately 9 mm tall) were roughened with population abundances of ants vary with biotic and abiotic sandpaper to allow ants to crawl easily over them, and were factors? and (3) What effects do hurricane-strength distur- not observed to be a barrier to any species. bances have on the ant populations occupying these small The baits were left out for 45 min, at which time the ant islands? species present at the baits were recorded. All surveys were conducted between 8:00 am and 6:00 pm. Negative inter- specific interactions sometimes resulted in workers of one METHODS species being excluded from within the Petri dishes, but not Study sites from the immediate vicinity of the baits. I recorded both species that were within Petri dishes (with direct access to This study was conducted on small islands in the central baits) and species around the perimeter that were prevented Exumas and near the northeast coast of Andros, Bahamas. In access to the Petri dishes by other species. Thus, all species the Exumas, the study area included the 13.5 km island that were attracted to the baits were sampled. The methods chain between O’Briens Cay to the north and Bitter Guana employed were exactly the same as those used in previous Cay to the south. The study area at Andros included the surveys (see Morrison, 1998a). small islands lying offshore the main island of Andros, Because relatively few ant species inhabit these islands, between Nicholls Town to the north and Staniard Creek to accurate field identification of most species was possible. the south (see maps in Morrison, 1997a). Specimens of morphologically similar congeners were col- All islands with a vegetated area <1000 m2 were lected and later identified in the lab to verify field identifi- surveyed, with the exception of ‘outer islands’ in the Exumas cations. Slight variations in coloration and eye size in that occurred in deep water and were mostly denuded by Brachymyrmex obscurior Forel indicate that more than one wave action. Islands with more than 1000 m2 of vegetated species may exist under this name, but the observed area were not surveyed because species with small popula- differences were not pronounced enough to allow separation tions could have been overlooked. into distinct morphospecies. Thus if the populations Ó Blackwell Science Ltd 2002, Journal of Biogeography, 29, 387–394 Bahamian ant island biogeography 389 identified as B. obscurior do contain cryptic species, turnover turnover, I conducted a series of t-tests. Four different island for this group may be greater than the values reported here. characteristics were evaluated in separate tests: vegetated Until this genus is thoroughly revised, however, this issue area, distance from the nearest large island, elevation and cannot be entirely resolved. Reference specimens have been number of plant species. All variables were log-transformed placed in the Bohart Museum of Entomology at the to normalize the distributions. For each characteristic, University of California at Davis. I tested whether islands on which turnover occurred were significantly different than islands on which turnover did not occur.